Interactive media systems may receive and process text information from an automatic speech recognition (ASR) module and retrieve information along with an appropriate response for the user's query. Such systems may also allow the user to retrieve information using a dialogue rather than a single transaction. However, whenever the ASR module misinterprets the user's speech, current interactive media systems tend to retrieve incorrect information since they do not know that the current query has an ASR error. Furthermore, since the user does not know when and where the ASR error occurred, he would need to repeat the entire dialogue, possibly multiple times, to get the correct information.